Agent runtimes in Go. Systems plumbing in Rust. Currently deep in RL and distributed systems.
| Project | What's interesting about it | |
|---|---|---|
| lucky-agent | Go |
An agent framework built around long-horizon planning and durable memory — the two things that break first once a task runs for hours instead of seconds. Docs → |
| kiro.rs | Rust |
Anthropic-compatible API layer with credential rotation, load balancing and prompt caching. |
| luckyagent-skill | Python |
Community skill library and reusable workflows for LuckyAgent. |
| agentcraft | JavaScript |
Multi-agent collaborative building game — agents negotiate, divide work, and build together. |
| yurika_blog | Rust Go TS |
Personal blog, full stack: axum API, a Go chat service, and a React front end. |
Smaller tools
| Project | ||
|---|---|---|
| Luckypay | Rust |
个体户的钱包 |
| LuckyForge | Go |
|
| bgm-cli | Go |
Bangumi from the terminal |
| Ghostype | AI completion plugin for Obsidian | |
| chatgpt2api-structured | Python |
response_format compatibility for structured output |
| Mini-Compiler | C++ |
Notes and solutions, kept in the open so the gaps are visible too.
- Distributed systems — MIT 6.824 · MIT 6.5840
- RL & LLM internals — APPO notes · Stanford CS336
- Fundamentals — MIT 6.1210 Algorithms · CS61A
- Reverse engineering — OpenSecurityTraining2
